This paper points out that the puzzling nature of the evidence concerning the relationship between teaching experience and teaching performance is due in part to the research methodology used to study this relationship. Cross-sectional data reflects three different phenomena: learning by doing, teachers' academic excellence, and influence of the job market on selection of teachers. It is explained how these phenomena influence the relationship between experience and performance observed in a cross-section of teachers. It is suggested that teachers' academic excellence has a marked effect on the estimated impact of learning by doing on teaching performance. (Author/JD)